The early identification and forecasting of epidemics using social media data is known as epidemic prediction using the Twitter dataset. Forecasting and identifying infectious diseases early are crucial to limit their impact on communities. Pandemics and other infectious diseases pose a major threat to public health. Due to their real-time and user-generated nature, social media sites like Twitter have emerged as important data sources for epidemic surveillance. However, manually analyzing huge amounts of Twitter data is difficult and time-consuming. This research intends to investigate how well Twitter data may be used to forecast epidemic outbreaks. A collection of tweets about the influenza epidemic was acquired, processed, and studied using various supervised machine learning techniques such as Support Vector Machine (SVM), Logistic Regression, Decision Tree, and Random Forest are examined. Text processing and Natural Language Processing (NLP) are essential components that focus on understanding and manipulating human language. This research also uses these NLP techniques to process the Twitter Data. On test data, the Random Forest model enhances previous model architectures and achieves the highest accuracy over 0.98 to classify epidemic-related activity into the four stages of early warning, peak, outbreak, and decay, followed by SVM, which is efficient.